Built-In Tools Overview

Powerful tools to enhance your AI applications with search and information retrieval capabilities

Our API offers several powerful built-in tools that enable advanced search and information retrieval capabilities for your AI applications.

Tools Overview

Think Tool

For logging thoughts without changing state

File Search Tool

For direct vector store searches

Agentic Search Tool

For AI-guided iterative searches

Feature	Think Tool	File Search Tool	Agentic Search Tool
Purpose	Document reasoning	Direct information retrieval	Complex research
Complexity	Simple	Moderate	High
Best for	Audit trails	Known-item searches	Research questions
Search iterations	None	Single	Multiple
Response time	Instant	Fast	Moderate

Built-In Tools Comparison

How Built-In Tools Work

Built-In Tools Data Flow

Choosing the Right Tool

• Think Tool: Use when you need to document reasoning steps or decision processes without changing state.
• File Search Tool: Ideal for direct, immediate results on simple queries when you know exactly what you’re looking for.
• Agentic Search Tool: Perfect for complex, multi-faceted questions requiring comprehensive information gathering and deep exploration.

Getting Started

Basic Configuration

Start with minimal configuration for each tool:

Think Tool

curl --location 'http://localhost:6644/v1/responses' \
--header 'Content-Type: application/json' \
--header 'Authorization: Bearer $YOUR_API_KEY' \
--data '{
    "model": "openai@gpt-4o",
    "tools": [
      {
        "type": "think"
      }
    ],
    "input": "Analyze the pros and cons of microservices architecture",
    "instructions": "Document your analysis process using the think tool."
}'

Advanced Options

For more control, each tool offers additional configuration parameters:

File Search Tool Options

• filters: Optional filters to narrow search results
• max_num_results: Maximum number of results to return (default: 20)

Agentic Search Tool Options

• max_num_results: Maximum number of final results to return
• max_iterations: Maximum number of search iterations
• seed_strategy: Strategy for initial seed queries
• alpha: Balance between vector and keyword search weights
• initial_seed_multiplier: Multiplier for initial seed generation
• enable_presence_penalty_tuning: Enable dynamic tuning of presence penalty
• enable_frequency_penalty_tuning: Enable dynamic tuning of frequency penalty
• enable_temperature_tuning: Enable dynamic tuning of temperature
• enable_top_p_tuning: Enable dynamic tuning of top p
• filters: Optional filters to narrow search results

For detailed configuration options and best practices for each tool, refer to the individual tool documentation pages.

Example Applications

• Customer Support: Use the Agentic Search Tool to research complex customer issues across your knowledge base.
• Legal Research: Find relevant precedents and statutes with the File Search Tool.
• Code Documentation: Document reasoning behind architectural decisions with the Think Tool.
• Medical Research: Use Agentic Search for comprehensive literature reviews on medical topics.

Performance Considerations

• File Search is faster but less thorough
• Agentic Search provides better results for complex questions but uses more resources
• For time-critical applications, set appropriate timeouts and consider using File Search

Ready to Get Started?

Explore the detailed documentation for each built-in tool to learn more about their capabilities and how to implement them in your applications.

Think ToolFile Search ToolAgentic Search Tool